Controversy continues over the significance of trans fatty acids in human nutrition, particularly concerning their negative effect on the plasma lipoprotein profile with its untoward implications for atherogenesis (Ascherio and Willet 1995, Judd et al. 1994, Katan et al. 1995, Willet et al. 1993). Several reports clearly demonstrate that modest intake of trans fatty acids can deleteriously affect lipoproteins by increasing LDL and decreasing HDL (Abbey and Nestel 1994, Almendingen et al. 1995, Judd et al. 1994, Katan et al. 1995) while inducing a rise in the atherogenic lipoprotein (a) [Lp(a)]⁴ (Mensink et al. 1992, Nestel et al. 1992) relative to the cis isomers. An outstanding question, however, is whether trans fatty acids are nutritionally better or worse in this regard than the dietary saturated fatty acids they were designed to replace in margarine, shortenings and in various frying and baking procedures.

The main fatty acids resulting from partial hydrogenation of vegetable oils are trans 18:1 and its various isomers [(n-8), (n-9) or (n-10)], but small quantities (<2%) of other cis-trans, trans-cis, or even trans-trans fatty acids are typically included in most partially hydrogenated oils. Most previous comparisons in humans have considered the metabolic effect of exchanging trans 18:1 for cis 18:1 rather than for the natural saturated fatty acids (12:0-18:0) that they physically mimic (Katan et al. 1995). Judd et al. (1994) included a saturated fat comparison and found that LDL cholesterol (LDL-C) was elevated similarly to 7% en from trans, but trans fatty acids depressed HDL cholesterol (HDL-C). A recent Norwegian study (Almendingen et al. 1995) fed butter as a saturated fat control for comparison with partially hydrogenated soybean oil (PHSBO) or fish oil (PHFO). Although both trans preparations elevated Lp(a), the butter diet was intermediate between the PHSBO and PHFO in its adverse effect on the lipoprotein profile, with PHFO eliciting significant elevations in LDL and the LDL/HDL ratio.

Because the major cholesterol-raising effect of saturated fats has been postulated to reside in their 12:0-16:0 fatty acids (Hayes 1995, Zock et al. 1994), our aim was to compare the specific effect of elaidic acid [trans 18:1(n-9)] on lipoproteins in normolipemic subjects following careful exchange for 12:0 + 14:0, 16:0 or cis 18:1.

MATERIALS AND METHODS

Table 1. Fatty acid composition of fat blends incorporated into diets [View Table]

The study used a double-blind randomized crossover design, and each dietary period lasted 4 wk while incorporating the same basic Malaysian foods, which are relatively low in fat. Using a 7-d rotating menu, the selected fats were readily incorporated into the foods as the sole cooking fat via the typical Malaysian stir-fry or deep-fry preparation of ingredients. Evaluation of the dietary fat sources indicated that added cooking oils typically provided approximately two-thirds of the total 26% dietary energy from fat. During the experimental periods a similar two-thirds substitution of all fats with the target fat blends was achieved, but at a slightly increased fat intake approximating 31.5% energy. Subjects were provided three meals (breakfast, lunch and high tea), which were prepared fresh each day by a special caterer. A dietitian strictly monitored the attendance of the volunteers at each meal and recorded their consumption patterns. Adherence to the preset menus and cooking oil allotments was monitored in the kitchen during all meal preparations. Because the subjects consumed their off-campus dinners with their families at home, they were provided with the appropriate cooking oil during each dietary period. The use of the oil in their homes was recorded in a diary; this record served as a compliance marker. In addition, subjects were randomly selected to provide a double portion of their home-cooked meal for analysis of fat content and fatty acid composition. This was composited with the three meals provided at the cafeteria and analyzed as previously described (Sundram et al. 1994 and 1995). At the end of each dietary period, the actual home use of the cooking oil was recorded. These methods of monitoring produced excellent compliance. Subjects were also monitored to ensure a relatively constant caloric intake throughout the study, which allowed for a stable BMI as determined by weekly body weight records. A questionnaire at the end of the trial revealed that the subjects were unable to identify the order, source or dietary fat being consumed.

As an index of compliance, the fatty acid composition of serum was measured in each subject at the end of each dietary period. A Perkin Elmer AutoSystem gas chromatogram (Perkin-Elmer, Norwalk, CT) fitted with a 100-m capillary column (Supelco SP2560, Bellefonte, PA) was used for the analysis.

RESULTS

Table 2. Daily nutrient intake during the habitual and experimental periods analyzed from double portions of 7-d menus1,2 [View Table]

Table 3. Serum fatty acid composition following dietary treatments1 [View Table]

Compared with both the cis 18:1-rich and 16:0-rich diets, the total plasma cholesterol value (Table 4) was elevated and similar for the trans-rich and 12:0 +14:0-rich diets (5.22 and 5.15 mmol/L, respectively) with LDL-C being slightly higher during trans intake (3.81 and 3.57 mmol/L, respectively). These two diets induced significantly higher TC and LDL-C than either the cis 18:1-rich or the 16:0-rich diet, which had lower and comparable values (TC: 4.78 vs. 4.85 mmol/L and LDL-C: 3.17 vs. 3.15 mmol/L, respectively). Striking reductions in HDL-C were noted during trans consumption (1.05 mmol/L), which was significantly lower than cis 18:1 (1.25 mmol/L) or 16:0 (1.26 mmol/L), which were in turn slightly higher than 12:0 + 14:0 (1.18 mmol/L) (Table 5). These differences produced substantial differences in the LDL/HDL ratio which was elevated significantly by trans 18:1 relative to cis 18:1 or 16:0. The 12:0 + 14:0 diet produced an intermediate LDL/HDL ratio. Triglyceride values were unaffected by type of dietary fat. The apolipoprotein concentrations confirmed the cholesterol values, i.e., apo A1 was depressed by the trans 18:1 diet, whereas apo B was elevated, both relative to the cis 18:1-rich and 16:0-rich diets. The apo B/apo A1 ratio, in turn, was significantly elevated by trans 18:1 consumption. Serum Lp(a) values were relatively low and normal in this population with about half the values remaining below 10 mg/dL throughout all four test periods. No value exceeded 40 mg/dL for any subject. Because of the 20- to 30-fold range in values, the nonrandom distribution was split into quartiles to determine whether the apparent increase during trans fatty acid consumption was related to the inherent Lp(a) concentration. Only trans consumption consistently increased Lp(a), about 35% on average, relative to 16:0-rich diet which elicited the least variable response. Again, compared with 16:0-rich diet, the greatest mass increase occurred in the highest quartile (about 5 mg/dL), but the greatest percentage increase (55%) occurred in the lowest quartile, in which the 12:0 + 14:0-rich diet had an even greater effect. A total of 19 of the 27 subjects had increased Lp(a) during trans intake, and the average increase for the 19 was about 40% relative to the 16:0-rich fat.

Table 4. Effect of dietary fat on plasma lipids, lipoproteins, serum apolipoprotein and Lp(a) after dietary treatment1 [View Table]

Table 5. Lp(a) response to changes in dietary fatty acids by quartiles [View Table]

DISCUSSION

These results affirm a striking difference between the effects of saturated and trans fatty acids on human lipoprotein metabolism, i.e., trans fatty acids depress HDL whereas SFA typically increase HDL, both generally in conjunction with an LDL increase. Furthermore, the exaggerated response to the selective inclusion of elaidic acid suggests that it may be the principal trans fatty acid evoking the deleterious effect. This negative effect of elaidic acid concurs with the observation that among six trans fatty acids (including vaccenic) detected in platelets from humans with coronary artery disease, only concentrations of elaidic and t18:1(n-8) correlated significantly with the degree of angiographically determined disease (Hodgson et al. 1996). Selective generation of elaidic acid for our experiment was achieved with a standard commercial catalyst, applying industrial procedures to produce a fat having a melting point and iodine value comparable to most available margarines.

On the other hand, elaidic acid probably is not the sole trans capable of exerting adverse lipoprotein effects. The previously cited Norwegian study (Almendingen et al. 1995) compared butter wiith PHSBO and PHFO, (both of which contributed about 8% en as trans). Butter exerted a more deleterious influence on TC and the LDL/HDL ratio than PHSBO, but was less detrimental than PHFO. Nevertheless, PHSBO had seven times more total trans than butter and three times more trans 18:1 isomers (undefined) than PHFO, whereas the latter fat contained two-thirds of its trans as very long-chain fatty acids. The Norwegian data are particularly noteworthy relative to the risk of coronary heart disease (CHD) when one couples the lipoprotein alterations (Almendingen et al. 1995) and adverse effects on thrombogenesis (Almendingen et al. 1996) with the fact that the highest quartile of trans intake in the Norwegian cohort of the EURAMIC study (Aro et al. 1995) experienced five times the risk of acute myocardial infarction compared with the lowest quartile of trans intake, all of which suggest a cause and effect relationship may exist. In any event, the metabolism of specific trans requires further study to ferret out the potentially objectionable trans isomers if any are to remain in the food supply.

The subjects in this study were exceptionally sensitive to the selective substitution of trans fatty acids for certain saturates (12:0 + 14:0 and 16:0), or monounsaturates (cis 18:1), demonstrating the greatest HDL-C decline and increase in the LDL/HDL ratio yet reported for humans consuming a trans fatty acid-rich fat (Abbey and Nestel 1994, Judd et al. 1994, Nestel et al. 1992). Other than the selective use of elaidic acid, reasons for the unusual sensitivity to trans fatty acids might include the dietary naiveté of the population. Aside from considerable exposure to saturates from 12:0 + 14:0 in coconut oil and 16:0 in palm oil, this population habitually ingests low amounts of total fat (26-30% en) and 18:2 (about 3% en), with minimal dietary cholesterol (<225 mg/d) and limited (<0.1% en) prior exposure to dietary trans (Ng et al. 1992, Sundram et al. 1994 and 1995). Furthermore, subjects were relatively lean with good BMI profiles (21-23 kg/m²). Population differences may be a contributing factor in light of a recent report that the plasma lipoprotein response to dietary saturated fat and cholesterol depends in part on ethnic background, with subjects of Afro-American and Asian descent being less sensitive to both factors than Caucasians (Fielding et al. 1995). It is possible that the inverse may be true for trans fatty acids, i.e., Malaysians may be more sensitive to trans than Caucasians. Whether this represents a true genetic difference or a subtle aspect of dietary ethnicity awaits further experimentation.

These data are provocative in their rejection of the hypothesis (tested herein) that regularly consumed saturated fatty acids (12:0, 14:0, 16:0) "are still less desirable" than trans fatty acids in our food supply (Clifton 1994). Although the public health implications are complicated by the fact that partially hydrogenated oils vary considerably in their composition and general application and presumably in their biological effect as well, one can no longer assume that the consumption of trans fatty acids is always preferable to the SFA they were designed to replace. Indeed, epidemiological data suggest that trans can be at least equal to (Kromhout et al. 1995), if not worse than saturates (Willet 1995) in terms of CHD risk, and they carry an as yet undefined risk for growth of the humans fetus (Koletzko 1992). At least one European country is committed to removing trans from the market place based on the negative association with CHD risk (Gillman et al. 1995), which our results would suggest may be greater than that produced by the most adverse saturated fatty acids.

A recent review of the epidemiological evidence concerning the CHD risk associated with trans intake (Kris-Etherton 1995) suggested that it was not possible to separate naturally occurring trans in ruminant fats, e.g., trans 16:1 and trans 18:1(n-7) (vaccenic acid), from trans fatty acids produced by hydrogenating vegetable oil. The present data clearly indict the trans fatty acids in hydrogenated vegetable oils, at least from the lipoprotein perspective, because essentially no dairy products and minimal ruminant meats were included in these typical Malaysian diets in which chicken and fish represent the main sources of animal protein and non-vegetable fats. The trans source was exclusively partially hydrogenated soybean oil and almost 80% elaidic acid by analysis. From that perspective, the specificity of our trans preparation differs somewhat from previous human studies in which a cluster of isomers more typical of commercial margarines was fed or the trans content was loosely characterized from food composition tables (Abbey and Nestel, 1994, Mensink and Katan 1990, Nestel et al. 1992). Clearly, elaidic acid-rich margarine adversely affects the LDL/HDL ratio and Lp(a) in humans.

An explanation for the rise in the LDL/HDL ratio was unexplored in the present study, but a similar shift during trans consumption has been attributed to increased cholesteryl ester transfer protein (CETP) activity, i.e., enhanced transfer of cholesteryl ester (CE) from HDL to lower density lipoproteins, including LDL. Specifically, Abbey and Nestel (1994) noted a 10% rise in CETP activity during margarine consumption. In individuals or species in which hepatic LDL receptors are encumbered (i.e., fully saturated or partially down-regulated), increased CE transfer from HDL might be expected to diminish the HDL-CE pool and overload the LDL-CE pool whenever LDL clearance was impaired. In cebus monkeys fed a cholesterol-free diet, in which LDL receptor activity and clearance of LDL are highly efficient, the same trans fat preparation fed in the present study was found to elevate CETP activity and depress HDL without altering the LDL-C pool size or LDL clearance (Khosla et al. 1996). As pointed out by others (Willet 1995, Zock et al. 1995), any dietary manipulation that increases the LDL/HDL ratio, particularly by increasing the absolute pool of LDL, bodes ill for CHD risk, not only from LDL deposition in arteries but also because an elevated LDL/HDL ratio has a negative effect on platelet aggregation and thrombogenesis (Ross 1993), an exceedingly deleterious aspect of atherogenesis (Hayes and Pronczuk 1996).

On the whole, Lp(a) values in these subjects were relatively low, possibly reflecting their historically low exposure to trans. Nonetheless, the trans fatty acid-induced rise in Lp(a) followed a pattern noted by others (Mensink et al. 1992, Nestel et al. 1992), i.e., the absolute increase associated with trans was greatest in individuals with the highest initial values, with the effect of trans being much less than the inherent 10-fold differences due to genetics. This adverse effect of trans on Lp(a) coupled with its untoward modulation of the plasma lipoproteins could trigger an increased risk for CHD in individuals continuously exposed to hydrogenated fats in their diets. In contrast, Clevidence et al. (1995) reported that trans monounsaturated fatty acids did not alter Lp(a) levels in their subjects when fed amounts reflecting typical American dietary intake levels.

As in previous studies of normolipemic individuals (Ng et al. 1992, Sundram et al. 1995), the effects of 16:0 and cis 18:1 on LDL and HDL were comparable, reaffirming the fact that dietary 16:0 need not raise TC when lipoprotein metabolism is unencumbered (Hayes 1995, Sundram et al. 1995). This would appear to include subjects with base-line LDL-cholesterol values 3.70 mmol/L.

ACKNOWLEDGMENTS